CA1 Contributes to Microcalcification and Tumourigenesis in Breast Cancer Yabing Zheng1, Bing Xu1, Yan Zhao1,Hegu1, Chang Li2, Yao Wang1 and Xiaotian Chang1*

Zheng et al. BMC Cancer (2015) 15:679 DOI 10.1186/s12885-015-1707-x

RESEARCH ARTICLE Open Access CA1 contributes to microcalcification and tumourigenesis in breast cancer Yabing Zheng1, Bing Xu1, Yan Zhao1,HeGu1, Chang Li2, Yao Wang1 and Xiaotian Chang1*

Abstract Background: Although mammary microcalcification is frequently observed and has been associated with poor survival in patients with breast cancer, the genesis of calcification remains unclear. Carbonic anhydrase I (CA1) has been shown to promote calcification by catalysing the hydration of CO2. This study aimed to determine whether CA1 was correlated with microcalcification and with other processes that are involved in breast cancer tumourigenesis. Methods: CA1 expression in breast cancer tissues and blood samples was detected using western blotting, real-time PCR, immunohistochemistry and ELISA. Calcification was induced in the cultured 4T1 cell line originating from mouse breast tumours, using ascorbic acid and β-glycerophosphate. Acetazolamide, a chemical inhibitor of CA1, was also added to the culture to determine the role of CA1 in calcification. The MCF-7 human breast cancer cell line was treated with anti-CA1 siRNA and was assessed using a CCK-8 cell proliferation assay, an annexin V cell apoptosis assay, transwell migration assay and a human breast cancer PCR array. The tag SNP rs725605, which is located in the CA1 locus, was genotyped using TaqMan® genotyping. Results: Increased CA1 expression was detected in samples of breast carcinoma tissues and blood obtained from patients with breast cancer. A total of 15.3 % of these blood samples exhibited a 2.1-fold or higher level of CA1 expression, compared to the average level of CA1 expression in samples from healthy controls. Following the induction of calcification of 4T1 cells, both the number of calcium-rich deposits and the expression of CA1 increased, whereas the calcification and CA1 expression were significantly supressed in the presence of acetazolamide. Increased migration and apoptosis were observed in MCF-7 cells that were treated with anti-CA1 siRNA. The PCR array detected up-regulation of the androgen receptor (AR) and down-regulation of X-box binding protein 1 (XBP1) in the treated MCF-7 cells. Significant differences in the allele and genotype frequencies of rs725605 were detected in the cohort of patients with breast cancer but not in other tumours. Conclusion: The results of this study suggested that CA1 is a potential oncogene and that it contributes to abnormal cell calcification, apoptosis and migration in breast cancer. Keywords: Breast cancer, Microcalcification, Tumourigenesis, Carbonic anhydrase I (CA1), Androgen receptor (AR), X-box binding protein 1 (XBP1)

Background mammary calcifications: calcium oxalate and hydroxy- Mammary microcalcification is frequently associated apatite [2, 3]. Hydroxyapatite is also a well-documented with poor survival, and it occurs in 30 to 50 % of breast component of bone, and the deposition of it in bone tis- cancer patients [1]. Although the diagnostic value of sue requires the coordinated expression of several bone microcalcification in patients with breast cancer is of matrix proteins, which are synthesised by cells of osteo- great importance, the genesis of this calcification blastic lineage [4]. There has been evidence indicating remains unclear. There are two distinct forms of that calcium carbonate is involved in initial bone formation, although mineral deposits are typically comprised – * Correspondence: [email protected] of calcium phosphate and not calcium carbonate [5 7]. 1Medical Research Center of Shandong Provincial Qianfoshan Hospital, CA1 (carbonic anhydrase 1) is a member of the carbonic Shandong University, Jingshi Road 16766, Jinan, Shandong 250014, P. R. China anhydrase (CA) family, and it catalyses the reversible Full list of author information is available at the end of the article

© 2015 Zheng et al. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zheng et al. BMC Cancer (2015) 15:679 Page 2 of 15

hydration and dehydration reactions of CO2/H2CO3 [8]. Methods In vitro assays have demonstrated that CA1 not only Tissue collection enhances hydration reactions but also promotes the for- In the present study, all of the solid tissue samples and mation of CaCO3 [9, 10]. In a recent study, we detected blood samples that were used were collected from increased expression of CA1 in Saos-2, a human osteo- patients of Shandong Provincial Qianfoshan Hospital sarcoma cell line, following the induction of calcification (Jinan, China). The tumour diagnoses were verified with ascorbic acid and β-glycerophosphate. Following using histological methods, and pathological categorisa- treatment with acetazolamide, an anti-CA1 drug, both tions were performed according to the World Health CA1 expression and the formation of mineralised nod- Organisation (WHO) classification system. All of the ules decreased [11]. Furthermore, overexpression of CA1 included patients signed informed consent forms, and was detected in the synovial tissues of patients with an- this study was approved by the ethics committee of kylosing spondylitis, a disease characterised by abnormal Shandong Provincial Qianfoshan Hospital (reference bone formation in the spinal and sacroiliac joints [12]. number 2013012). All research involving human subjects The gene that encodes CA1 is susceptible to ankylosing (including human material or human data) that is spondylitis [11], and transgenic mice with overexpres- reported in the manuscript was conducted in compli- sion of CA1 exhibited bone fusion in their paws and ance with the Helsinki Declaration. spines [13]. These results suggest that CA1 might play essential roles in bio-mineralisation and new bone SNP selection and Taqman genotyping formation. The tag SNP rs725605 was identified by searching the In the present study, we hypothesised that the up- HapMap database. This tag SNP was selected on the regulation of CA1 in breast tumours stimulates calcium basis of linkage disequilibrium patterns that were ob- precipitation, similar to what occurs in bone tissue. To served in samples obtained from the Han Chinese popu- test this hypothesis, calcification was induced in 4T1 lation of Beijing, who were genotyped as part of the cells, which originated from a murine mammary adenocar- International HapMap Project. Illumina microarray and cinoma. Cox et al. recently reported a novel in vitro model Taqman genotyping assays confirmed the location of tag of mammary mineralisation using 4T1 cells [14, 15]. Using SNP rs725605 in the CA1 locus [11]. this model, we investigated the involvement of CA1 in To determine the potential associations between this bio-mineralisation and calcification. Additionally, the tag SNP and various cancer risks, genotyping was effects of CA1 on cell proliferation and cell apoptosis performed using TaqMan® technology in cohorts of were investigated in anti-CA1 siRNA-treated MCF-7 patients who had breast cancer (n = 285, 285 women, cells that originated from human breast cancer, and the mean age = 47.65 years old), colon cancer (n = 145, 55 pathogenic pathway was analysed using a PCR array women, mean age = 54.13 years old), oesophageal cancer that contained breast cancer genes. We also used a Taq- (n = 285, 40 women, mean age = 61.20 years old), cer- man genotyping method to determine the correlations vical cancer (n = 190, 190 women, mean age = 52.75 years between common polymorphisms in the CA1 encoding old), liver cancer (n = 190, 42 women, mean age = gene and breast cancer. In a previous study, we genotyped 54.05 years old), lung cancer (n = 190, 56 women, mean 13 different tag SNPs in the CA1 gene to determine age = 58.17 years old), gastric cancer (n =285,71women, whether a potential association existed between the CA1 mean age = 56.83 years old) or rectal cancer (n =137,50 gene and ankylosing spondylitis, using a custom-designed women, mean age = 54.61 years old), as well as in healthy Illumina 96-SNP VeraCode microarray (Illumina). We controls (n = 285, 71 women, mean age = 38.42 years old). subsequently performed Taqman genotyping of the tag Blood samples were collected and stored in Monovette SNPs that exhibited significant associations with the tubes containing 3.8 % sodium citrate. disease in a large cohort of the patients. Both microarray Genomic DNA was extracted from whole blood samples and Taqman genotyping demonstrated that the allele and with an Omega E-Z 96 Blood DNA kit (Omega, USA) gene frequencies of rs725605 were statistically significantly according to the manufacturer’s protocol. Genotyping associated with patients who had ankylosing spondylitis assays were run on a ViiA 7 DX (Life Technology) and [11]. At that time, we tested five SNPs by Taqman geno- were evaluated according to the manufacturer’sinstruc- typing, and only rs725605 produced favourable results. tions. Each reaction was performed in a total volume of Thus, we continued to genotype this tag SNP to deter- 10 μl using the following amplification protocol: denatur- mine whether the CA1 gene had possible associations ation at 95 °C for 10 min, followed by 50 cycles of with the various types of tumours that were examined in denaturation at 95 °C for 15 s and finishing with annealing the current study. Furthermore, we examined the CA1 and extension at 60 °C for 1 min. The genotype of each expression in breast tumour tissues and blood samples sample was determined by measuring allele-specific fluor- from patients. escence using Taqman Genotyper software, version 1.2 Zheng et al. BMC Cancer (2015) 15:679 Page 3 of 15

(Life Technology). Duplicate samples and negative controls tissue samples. Clinical data, including the age, sex, clin- were included to verify the accuracy of the genotyping. ical pathological diagnosis and origin of every participant, Genotyping quality was evaluated using a detailed qual- were provided by the manufacturer. Tissue sections were ity control procedure that consisted of a >95 % successful deparaffinised and rehydrated using standard procedures. call rate, duplicate calling of genotypes, internal positive Prior to the application of the antibodies, the tissue control samples and Hardy-Weinberg Equilibrium (HWE) sections were heated at 95 °C for 10 min in citrate buffer testing. SNPs were analysed for associations by comparing solution (Sigma) to facilitate antigen recovery, followed by the MAFs (minor allele frequencies) between cases and incubation with an endogenous peroxidase inhibitor controls. Dominant and recessive models were considered (MaixinBio, China) for 30 min at room temperature. After with respect to the minor allele. Associations of SNPs with washing with PBS buffer, the section was incubated with diseases were evaluated using odds ratios (ORs) with 95 % anti-CA1 overnight at 4 °C. The CA1 antibody was pre- confidence intervals (CIs). Fisher’s exact test was used pared as described above. The immunoreaction was proc- for comparisons between categorical variables. p values essed using the UltraSensitive TM S-P Kit (Maixin-Bio, less than 0.05 were considered statistically significant. China) according to the manufacturer’s instructions. The Genotypic association was assessed using SHEsis soft- immune-reactive signal was visualised using DAB sub- ware [16]. Multiple-test correction, including genomic- strate, which stains target proteins yellow. Cell structures control correction and Bonferroni’s single-step correc- were counterstained with haematoxylin. tion, was performed using Plink software, version 1.07 (http://pngu.mgh.harvard.edu/purcell/plink/). Real-time PCR Western blot analysis Breast tissues were collected during galactophore opera- Breast tissues were collected during galactophore opera- tions that were performed on patients with breast cancer tions from patients with breast cancer (n = 7, 7 women; (n = 12, 12 women; 31–68 years old, mean 51) or breast 34–64 years old, mean 52) or breast fibroadenoma (n =7, fibroadenoma (n = 12, 12 women; 12–59 years old, mean 7 female subjects; 17–59 years old, mean 35). Two 33). Total RNA extracts from breast cancer and breast hundred micrograms of each of the breast cancer tumour fibroadenoma tissues were reverse-transcribed using an and breast fibroadenoma samples were individually homo- RNA PCR Kit (TaKaRa). Real-time PCR was conducted genised in Cell Lysis Solution (Sigma) and were centri- using a ViiA 7 DX. The relative levels of mRNA expression fuged at 12,000 × g for 30 min at 4 °C. The supernatants were calculated using a comparative threshold cycle (Ct) were collected after centrifugation, and protein concentra- method. The relative expression levels of the target gene tions were determined using a BCA protein assay kit were normalised to the relative levels of GAPDH mRNA. (Pierce). In total, 30 μg of protein were loaded and sepa- The primer sequences that were used for the amplification rated by sodium dodecyl sulphate-polyacrylamide gel elec- of CA1 and GAPDH are described in Table 1. The reaction trophoresis (SDS-PAGE), transferred onto a polyvinylidene conditions were as follows: 10 s at 65 °C, followed by membrane and probed with an anti-human carbonic anhy- 40 cycles of 5 s at 60 °C and 10 s at 72 °C and then 30 s at drase 1 antibody (1:1000; Abcam). The antibody was pre- 65 °C. The experiment was performed in triplicate, and the pared by immunising a goat with carbonic anhydrase I that PCR products were confirmed using a melting curve ana- was extracted from human erythrocytes. The manufacturer lysis. Additionally, real-time PCR was used to verify the confirmed that no cross-reactivity existed with other PCR array results. The primer sequences that were used carbonic anhydrases. The membranes were subse- for the amplification of AR (androgen receptor), GSTP1 quently rinsed with washing solution and incubated (glutathione S-transferase pi 1), PTGS2 (prostaglandin- with sheep anti-goat IgG conjugated with peroxidase endoperoxide synthase 2 [prostaglandin G/H synthase and (1:500; Sigma-Aldrich) for 30 min. Following a washing step, the immunosignal was detected using an Enhanced Table 1 Prime sequences for real time PCR (5'- > 3') Chemiluminescence (ECL) Plus kit (Millipore) and was Forward primer Reverse primer quantified using ImageQuant software, version 5.2 (GE CA1 GCTACAGGCTCTTTCAGTT GACTCCATCCACTGTATGTT Healthcare). Another membrane was prepared using the same protocol and was probed with an anti-GADPH anti- GAPDH ACCACAGTCCATGCCATCAC TCCACCACCCTGTTGCTGTA body (Santa Cruz) to normalise sample loading. AR CCAGGGACCATGTTTTGCC CGAAGACGACAAGATGGACAA GSTP1 CCCTACACCGTGGTCTATTTCC CAGGAGGCTTTGAGTGAGC Immunohistochemistry PTGS2 TAAGTGCGATTGTACCCGGAC TTTGTAGCCATAGTCAGCATTGT Tissue array slides were obtained commercially from SNAI2 TGTGACAAGGAATATGTGAGCC TGAGCCCTCAGATTTGACCTG Alenabio (China). The array slides contained 40 different XBP1 CCCTCCAGAACATCTCCCCAT ACATGACTGGGTCCAAGTTGT invasive breast ductal carcinoma and 8 different normal Zheng et al. BMC Cancer (2015) 15:679 Page 4 of 15

cyclooxygenase]), SNAI2 (snail family zinc finger 2) and Alizarin Red-S staining and quantification of XBP1 (X-box binding protein 1) are also shown in bio-mineralisation Table 1. The cultured cells were washed with phosphate-buffered saline (PBS, Solabio) and were stained with 0.5 % (w/v) Alizarin Red-S (AR-S, Solabio) in PBS (pH = 5.0) for ELISA 30 min at room temperature. After being washed four Blood samples were collected from patients with breast times with PBS, the stained cells were photographed. cancer (n = 92, 90 women; 22–78 years old, mean 53) The cultured cells were then de-stained with 10 % (w/v) and healthy volunteers (n = 84, 82 women; 23–77 years cetylpyridinium chloride (Biobasic) in PBS (pH = 7.0) old, mean 52). The blood samples were centrifuged at for 1 h at room temperature. The AR-S concentration 3000 × g for 10 min at 4 °C to remove debris. The was determined by measuring absorbance at 562 nm. samples were diluted 5-fold with 0.05 M carbonate- The protocol was based on previously published studies bicarbonate buffer (pH 9.6) and were used to coat 96- [11, 19]. well ELISA microplates (Costar), followed by overnight Expression levels of bone sialoprotein (BSP) and osteo- incubation at 4 °C. After a brief washing with PBS con- calcin (OCN), two bone matrix proteins, were deter- taining 0.1 % Tween-20 (PBST), the plates were blocked mined using real-time PCR. The expression of CA1 in with 5 % non-fat dry milk for 1 h at room temperature. the treated 4T1 cells was determined using real-time The anti-CA1 antibody was diluted 1000-fold with PCR and western blot analysis. PBST and was added to the plate, and the plate was then incubated for 2 h at room temperature. After Inhibiting CA1 expression with siRNAs in MCF-7 cells washing with PBST, a 1000-fold dilution of HRP- The MCF-7 cell line, which originated from human conjugated anti-goat total IgG (Sigma) was added, and breast cancer mammary epithelium, was cultured in the plate was incubated at room temperature for 1 h. regular growth media that consisted of RPMI-1640, 10 % The plates were washed three times and developed FBS and 1 % penicillin/streptomycin. The cell line was using tetramethyl benzidine (TMB, Sigma) as a sub- purchased from the American Tissue Culture Collection. strate.O.D.wasmeasuredwithamicroplatereaderat All of the cell culture reagents were purchased from 450 nm. Sigma-Aldrich. siRNA oligonucleotides targeting CA1 (target mRNA sequence: 5′-TCTACTCTCCTTCCTT Cell culture and induction of bio-mineralisation CAT -3′) were designed and synthesised by RIBOBIO. The 4T1 cell line, which originated from a murine The cultured MCF-7 cells were transfected with the siR- mammary adenocarcinoma, was maintained in regular NAs at 20 nM using HiPerFect transfection reagent growth media that consisted of low glucose DMEM, (Qiagen), according to the manufacturer’s protocol. The 10 % FBS and 1 % penicillin/streptomycin. The cell line cells were harvested for analysis 72 h after the transfec- was purchased from the American Tissue Culture tion. Parallel experiments with AllStars siRNAs, which Collection. All of the cell culture reagents were pur- was provided with the kit, were used as a negative con- chased from Sigma-Aldrich. The 4T1 cells were seeded trol. The AllStars siRNA has no homology to any known into 6-well culture plates at a density of 105 cells/well mammalian gene and therefore should not affect any and were induced for calcification with an osteogenic gene expression. Additionally, the MCF-7 cells that were cocktail (OC) (50 g/ml ascorbic acid, 10 mM β- transfected with HiPerFect transfection reagent were glycerophosphate) the following day. The cells were also used as a control in the experiment. maintained in an atmosphere of 37 °C and 5 % CO2, and half of the medium was replenished every 3 days Cell proliferation assay until the 28th day. This protocol design was based on a MCF-7 cells were seeded into 96-well culture plates and study by Cox et al. [14, 15]. incubated until they reached 80 % confluence. The To verify the effects of CA1 expression on bio- culture was then treated with anti-CA1 siRNA and incu- mineralisation, acetazolamide, a chemical inhibitor of bated for 24–72 h at 37 °C in 5 % CO2. Following the CA1, was used to treat the 4T1 cells. Acetazolamide addition of 10 μl of Cell Counting Kit-8 (CCK-8, (Sigma) was dissolved in dimethyl sulphoxide (DMSO, Dojindo) solution to each well, the cells were incubated Solabio) at a concentration of 1 % and was added to cell for an additional 4 h. Absorbance was measured at 450 nm cultures that were grown to 70 % confluence in OC. The with a spectrophotometer (Spectramax 190; Molecular cultures were grown in the presence of acetazolamide at Devices). Growth curves were generated from the average a final concentration of 1 mM. The 4T1 cells were also values of five wells per group. cultured in OC with 1 % DMSO as a control. The proto- The proliferation of the 4T1 cells that were cultured col was based on a study by Hall et al. [17, 18]. with OC was measured using the same protocol. Zheng et al. BMC Cancer (2015) 15:679 Page 5 of 15

Cell apoptosis assay 10 min at room temperature prior to washing with Apoptosis in the siRNA-treated MCF-7 cells was ana- water. The cells that had migrated to the lower side of lysed using flow cytometry (FACSAria II, BD Biosci- the filter were counted with an inverted fluorescence ences). The cultured cells were washed twice with PBS microscope. and were resuspended in binding buffer at a concentration of 1 × 106 cells/ml. The cell suspensions (1 × 105 PCR arrays and data verification cells/100 μl) were transferred into 5 ml culture tubes, A Human Breast Cancer RT2 Profiler™ PCR Array (Qiagen) and 5 μl of annexin V-phycoerythrin (eBioscience) and was used to screen for breast tumour genes that were 5 μl of 7-amino-actinomycin (eBioscience) were then affected by CA1 expression. The breast cancer pathway added. The cells were gently vortexed and incubated at PCR array profiled the expression of 84 different key room temperature in the dark for 15 min. Subse- genes that are commonly involved in the dysregulation quently, another 400 μl of aliquot of binding buffer of signal transduction and other normal biological were added. Flow cytometry was performed within 4 h processes during breast carcinogenesis. PCR array ana- of staining. lysis was conducted using a ViiA7 DX according to the manufacturer’s instructions. The procedure begins Transwell migration assay with the conversion of experimental RNA samples into Transwell inserts (8.0 μm pore size) with a polycarbon- first-strand cDNA using a RT2 First Strand Kit. Subse- ate filter (Costar®) were used to examine the effects of quently, the cDNA is mixed with an appropriate RT2 CA1 on cell migration. siRNA-treated MCF-7 cells (5 × SYBR Green Mastermix. This mixture is then ali- 104 cells/200 μl) were suspended in serum-free media quoted into the wells of the RT2 Profiler PCR Array. and added to the upper chamber, and 500 μlof Real-time PCR detection was performed under the completeDMEMmediawereaddedtothelowercham- following thermal cycling conditions: 95 °C for 10 min, ber. Following incubation for 24 h, the filter was 40 cycles of 95 °C for 15 s and 1 cycle of 60 °C for immersed in methanol for 15 min at room temperature 1 min. Relative expressions were determined using the and then was treated with 0.25 % crystal violet stain for data that were generated with the real-time cycler and

Table 2 Taqman genotyping (rs725605-CA1) result (control n = 288) SNP identity-gene Breast cancer Rectal carcinoma Colon carcinoma Cervical carcinoma Esophageal Gastric Liver cancer Lung n = 278 n = 132 n = 140 n = 179 carcinoma carcinoma n = 186 carcinoma n = 259 n = 263 n = 181 Allele A G A G A G A G A G A G A G A G Case (freq) 220 (0.396) 118 (0.447) 113 (0.404) 163 (0.455) 252 (0.486) 232 (0.441) 162 (0.435) 174 (0.481) 336 (0.604) 146 (0.553) 167 (0.596) 195 (0.545) 266 (0.514) 294 (0.559) 210 (0.565) 188 (0.519) Control (freq) 270 (0.475) 270 (0.475) 270 (0.475) 270 (0.475) 270 (0.475) 270 (0.475) 270 (0.475) 270 (0.475) 298 (0.525) 298 (0.525) 298 (0.525) 298 (0.525) 298 (0.525) 298 (0.525) 298 (0.525) 298 (0.525) Odds Ratio (% 95 CI) 0.722663 0.892034 0.746817 0.922583 1.045614 0.870950 0.851429 1.021513 (0.570361~ (0.665320~ (0.558798~ (0.707594~ (0.823876~ (0.686340~ (0.654582~ (0.784591~ 0.915634) 1.196004) 0.998101) 1.202893) 1.327031) 1.105216) 1.107471) 1.329979) Fisher’s p value 0.0071 0.44502 0.0483 0.55165 0.714 0.255551 0.230431 0.874386 Genotype A/A A/A A/A A/A A/A A/A A/A A/A A/G A/G A/G A/G A/G A/G A/G A/G G/G G/G G/G G/G G/G G/G G/G G/G Case (freq) 49 (0.176) 23 (0.174) 21 (0.150) 38 (0.212) 51 (0.197) 53 (0.202) 33 (0.177) 41 (0.227) 122 (0.439) 72 (0.545) 71 (0.507) 87 (0.486) 150 (0.579) 126 (0.479) 96 (0.516) 92 (0.508) 107 (0.385) 37 (0.280) 48 (0.343) 54 (0.302) 58 (0.224) 84 (0.319) 57 (0.306) 48 (0.265) Control (freq) 67 (0.236) 67 (0.236) 67 (0.236) 67 (0.236) 67 (0.236) 67 (0.236) 67 (0.236) 67 (0.236) 136 (0.479) 136 (0.479) 136 (0.479) 136 (0.479) 136 (0.479) 136 (0.479) 136 (0.479) 136 (0.479) 81 (0.285) 81 (0.285) 81 (0.285) 81 (0.285) 81 (0.285) 81 (0.285) 81 (0.285) 81 (0.285) Fisher’s p value 0.029 0.30265 0.1048 0.825315 0.063 0.53125 0.317314 0.820064 HWE for case (df = 1) 0.1697 0.23526 0.527 0.787951 0.01 0.645998 0.497597 0.807632 HWE for control (df = 1) 0.5011 0.50114 0.5011 0.501122 0.501 0.501136 0.501122 0.501122 Zheng et al. BMC Cancer (2015) 15:679 Page 6 of 15

Fig. 1 CA1 expression in breast cancer and breast fibroadenoma. a Visualisation of CA1 expression and GAPDH expression in breast cancer tissues using western blot analysis. b Visualisation of CA1 expression and GAPDH expression in breast fibroadenoma using western blot analysis. c For quantification purposes, the level of CA1 protein expression was normalised against GAPDH expression in each of the tissues. d Quantifying the expression of CA1 mRNA in breast cancer tissue and breast fibroadenoma using real-time PCR; CA1 expression was normalised against GAPDH expression in each of the tissues. The expression levels are depicted as the means ± SEMs. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001

the ΔΔCT method. Another PCR array experiment performing real-time PCR on RNA extracts from anti- was prepared with total RNA extracted from the cells CA1 siRNA-treated MCF-7 cells. with treatment of Allstars siRNA. The transcription levels of the target genes were normalised to their Statistical analyses corresponding expression levels in the cells that were Statistical analyses of the data were performed using SPSS treated with Allstars siRNA. Data analysis was per- software, version 16 (SPSS). Multiple comparisons were formed using Web-based RT2 Profiler PCR Array Data conducted using ANOVA. The t-test was conducted to Analysis software (http://pcrdataanalysis.sabiosciences.com/ assess significant differences between two groups. P values pcr/arrayanalysis.php, SA Biosciences). Fold-changes in less than 0.05 were considered significant. The data are gene expression were calculated and expressed as log- presented as standard deviations. normalised ratios of the siRNA treated cells/controls. According to the instructions of the manufacturer, genes Results with at least a 4-fold changes in expression were consid- Genotyping rs725605 in cohorts of patients with tumours ered to be biologically significant in the study. The Using the TaqMan method, genotyping of the tag SNP results of the PCR array assay were confirmed by rs725605 was performed in samples obtained from

Fig. 2 Immunolocalisation of CA1 expression in breast cancer tissues. a CA1 expression was significantly increased in breast cancer tissue. b CA1 expression was very low in normal breast tissue. The immune signals were located in tumour cells. Original magnification: ×20 Zheng et al. BMC Cancer (2015) 15:679 Page 7 of 15

Fig. 3 The detection of CA1 levels in patient blood samples using ELISA. The levels of CA1 are represented by O.D. values, which were measured at 405 nm. The results are expressed as the means ± SEMs. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001 cohorts of patients with breast cancer, colon cancer, oesophageal cancer, cervical cancer, liver cancer, lung cancer, gastric cancer, or rectal cancer, in addition to healthy controls. The allelic and gene frequencies of the SNP did not deviate from HWE in any of the cohorts. The allele frequency (OR = 0.722663, 95 % CI = 0.570361– 0.915634, p = 0.007104) and gene frequency (p = 0.029009) of SNP rs725605 were both statistically associated with breast cancer. Following multivariate logistic regression analysis using Plink software, version 1.07, the correlations between the SNP frequencies and breast cancer were found to be statistically significant. The genotyping data did not indicate significant differences in the allelic or genotypic frequencies of rs725605 (p > 0.05) between patients with colon cancer, oesophageal cancer, cervical cancer, liver cancer, lung cancer, gastric cancer or rectal cancer and healthy controls. The results are shown in Table 2. SNP rs725605 is located in the intron 1 region of thegenethatencodesCA1.

Expression of CA1 in breast cancer tissues Western blot analysis was performed to detect the expression of CA1 in breast tumour tissue and breast fibroadenoma. Using GAPDH as a reference, western blotting revealed that the expression of CA1 was significantly increased in the breast cancer samples, compared to the breast fibroadenoma samples (p = 0.008). Six of seven of the breast cancer samples exhibited increased levels of expression. Only two of seven of the breast fibroadenoma samples exhibited relatively high CA1 expression. These results are shown in Fig. 1. Fig. 4 Inducing calcification in 4T1 cells with ascorbic acid and β-glycerophosphate. 4T1 cells were incubated in normal media (control) The transcription of CA1 was also examined using real- and an osteogenic cocktail (OC). a The cell cultures were stained with time PCR. Similar to the western blotting results above, AR-S to detect calcium nodules and were photographed at the original the breast cancer samples (n = 12) exhibited a higher magnification and at ×100 magnification. b The cell cultures were degree of CA1 mRNA expression than the breast fibro- stained with cetylpyridinium chloride and were quantified by measuring p p adenoma tissue samples (n =12; p = 0.0000525). These the absorbance at 562 nm. * indicates <0.05,**indicates <0.01, *** indicates p < 0.001 results are shown in Fig. 1. Zheng et al. BMC Cancer (2015) 15:679 Page 8 of 15

Fig. 5 Analysis of cell proliferation using CCK-8 assay. 4T1 cells were incubated in either normal culture media (control) or osteogenic medium (OC). The numbers of viable cells are represented by O.D. values, which were measured at 492 nm. The error bars represent the standard deviation from five repeated measurements. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p <0.001

Immunohistochemistry was performed to visualise CA1 levels in blood samples obtained from patients with CA1 expression in a panel of breast tumour tissues. breast cancer There was significant expression of CA1 in 39 of 40 Sandwich ELISA was used to measure the levels of CA1 (97.5 %) of the invasive breast ductal carcinomas. There in the blood samples that were obtained from patients was no obvious expression of CA1 in normal breast with breast cancer. There were significantly increased tissue except for in select mesenchymal-like cells and select endothelial cells. The immunosignals were local- ised in the cytoplasm of breast tumour cells. The immunohistochemical results are shown in Fig. 2.

Fig. 6 Detecting the mRNA expression of BSP and OCN in 4T1 cells using real-time PCR analysis. 4T1 cells were incubated with an Fig. 7 CA1 expression in the calcified 4T1 cells. a CA1 transcription osteogenic cocktail (OC). These cells, which were cultured in was detected using real-time PCR in 4T1 cells that were treated with normal media, were used as a control. a BSP mRNA expression. osteogenic cocktail (OC). b CA1 protein expression in 4T1 cells b OCN mRNA expression. The expression was normalised to the treated with OC was detected using western blot analysis. c The expression of GAPDH. * indicates p < 0.05, ** indicates p < 0.01, protein expression level was normalised to the expression of GAPDH. *** indicates p < 0.001 * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001 Zheng et al. BMC Cancer (2015) 15:679 Page 9 of 15

levels of CA1 in the samples that were obtained from Bio-mineralisation, cell proliferation and expression of breast cancer patients (OD average = 0.91 ± 0.46) compared CA1, BSP and OCN in 4T1 cells to the samples that were obtained from healthy controls Following the induction of bio-mineralisation with OC, (OD average = 0.55 ± 0.3). Fourteen of 92 (15.22 %) of the numerous calcium deposits were observed in cultured samples that were obtained from patients with breast 4T1 cells after incubations of 7, 14, 21 and 28 days. The cancer exhibited two-fold or greater expression of CA1, absorbance of cetylpyridinium chloride was 1.64-fold compared to the average level of expression that was mea- higher in 4T1 cells that were incubated with OC for sured in the samples that were obtained from healthy 7 days than in cells that were cultured without induction − controls. Only one sample from the healthy control group (p = 4.32 10 6), and it was 2.57-fold higher at 14 days − (1.19 %) exhibited a higher level of CA1 expression than (p = 2.04 × 10 12), 4.46-fold higher at 21 days (p = 6.05 × − − the average level in the healthy samples. The serum levels 10 16), and 6-fold higher at 28 days (p = 7.87 × 10 13). of CA1 were significantly different between breast cancer These results are shown in Fig. 4. − patients and healthy controls (p =9.18×10 5). The ELISA 4T1 cells were incubated with OC for 2 and 3 days, results are shown in Fig. 3. and viable cells were detected using CCK-8 assay. There

Fig. 8 Inhibiting calcification in 4T1 cells with acetazolamide. The 4T1 cells were incubated with osteogenic cocktail (OC) supplemented with 1 mM acetazolamide (AZ). a The 4T1 cells cultured with OC supplemented with DMSO in the presence or absence of AZ were stained with AR-S and were photographed at the original magnification and at ×100 magnification. b The 4T1 cells, cultured with OC supplemented with DMSO in the presence or absence of AZ, were stained with cetylpyridinium chloride and were quantified by measuring their absorbance at 562 nm. c CA1 transcription in 4T1 cells was detected by real-time PCR. d CA1 protein expression in 4T1 cells was detected using western blot analysis. e The protein expression level was normalised to the expression of GAPDH. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001 Zheng et al. BMC Cancer (2015) 15:679 Page 10 of 15

were no significant differences in the numbers of viable cells that were detected between the 4T1 cells that were cultured in OC and those that were cultured in normal media at either 2 days (p = 0.875) or 3 days (p = 0.116). These results are shown in Fig. 5. − The level of BSP mRNA was 4.68-fold (p = 2.57 × 10 8) higher in the OC-stimulated 4T1 cells than in the cells without stimulation, and OCN increased by 5.67-fold − (p =3.01×10 5) when 4T1 cells were incubated in OC for 28 days. These results are shown in Fig. 6. The increased expression of BSP and OCN indicated that the 4T1 cells that were incubated in OC underwent ossification. The transcription and translation of CA1 in the OC- stimulated cells were measured using real-time PCR and western blotting, respectively. The level of CA1 mRNA was significantly elevated in the OC-stimulated cells compared with the controls (p = 0.00752) after 28 days in culture. The translation of CA1 in the OC-stimulated cells was also significantly increased compared with the controls (p = 0.0295). These results are shown in Fig. 7. To identify the essential role of CA1 for calcification in breast cancer cells, acetazolamide, a chemical inhibitor of CA1, was added into the culture medium when 4T1 cells were induced to produce calcification with OC. Compared with 4T1 cells that were treated with OC in the presence of DMSO, the formation of calcium nodules was significantly decreased in the cells cultured with OC in presence of DMSO and acetazolamide for 14, 21 and 28 days. Because acetazolamide was dissolved only in DMSO, the cells cultured with OC supplemented Fig. 9 CA1 expression in anti-CA1 siRNA-treated MCF-7 cells. a The with DMSO were used as a control in the experiment. expression of CA1 mRNA was detected using real-time PCR. b The The absorbance of cetylpyridinium chloride was also level of CA1 protein expression was detected using western blot considerably decreased in 4T1 cells that were treated analysis. c The level of CA1 protein expression was normalised against with acetazolamide for 14 days (p = 0.038), 21 days the expression of GAPDH. The cells without siRNA treatment were p p used as controls, and the cells treated with Allstars siRNA were ( = 0.012) and 28 days ( = 0.007). This experiment used as negative controls. * indicates p < 0.05, ** indicates p < 0.01, was repeated three times, obtaining similar results. *** indicates p < 0.001 TheresultsareshowninFig.8.

Suppressing CA1 expression in MCF-7 cells using siRNA The proliferation of MCF-7 cells that were treated MCF-7 cells were transfected with siRNA targeted to with anti-CA1 siRNA was measured using CCK-8 assay. CA1. The transcription and translation of CA1 in the No significant change in cell proliferation was detected treated cells were measured using real-time PCR and in the siRNA-treated MCF-7 cells, compared to the western blotting, respectively. Following the anti-CA1 cells that were treated with the negative siRNA control siRNA treatment, both the transcription (p = 0.004) and (p = 0.981) and the cells that were treated only with the translation (p = 0.0003) of CA1 were considerably HiPerFect transfection reagent (p = 0.801). These re- reduced in the siRNA-treated MCF-7 cells, compared sults, which are shown in Fig. 10, indicated that CA1 with the cells that were treated with the Allstars siRNA did not affect MCF-7 cell proliferation. negative control and the cells that did not receive siRNA The effect of CA1 on apoptosis in MCF-7 cells was treatment. Neither the control nor negative control determined using flow cytometric analysis with annexin siRNA led to a significant change in the expression of V/PE and 7-AAD double staining. Compared with the CA1 mRNA, indicating the specificity of the RNA inter- negative control (5.57 ± 0.15 %), the number of apoptotic ference with regard to CA1 expression in this experi- cells was significantly increased in the anti-CA1 siRNA- ment. These results are shown in Fig. 9. treated cells (p = 9.14E-8). No significant differences Zheng et al. BMC Cancer (2015) 15:679 Page 11 of 15

Fig. 10 Proliferation of MCF-7 cells that were treated with anti-CA1 siRNA, as measured by CCK-8 assay. Viable cell numbers are represented by an O.D. value that was measured at 405 nm. The cells without siRNA treatment were used as controls, and the cells treated with Allstars siRNA were used as negative controls. *indicates p < 0.05, **indicates p < 0.01, ***indicates p < 0.001 were detected between the control and negative control Discussion groups (p = 0.547). These results, shown in Fig. 11, Despite the diagnostic and potential prognostic value of demonstrate that down-regulation of CA1 expression microcalcification, the mechanisms underlying the forma- could induce apoptosis in MCF-7 cells. tion and functional role of microcalcification in breast MCF-7 cell migration was examined using a 2- cancer progression remain unclear. In the present study, compartment transwell system. MCF-7 cell migration in- we genotyped Tag SNP rs725605 in cohorts of patients creased significantly when CA1 expression was suppressed with various types of tumours and found that this tag by anti-CA1 siRNA (p < 0.001). No significant difference SNP, which is located in the intron 1 region of the gene in migration was observed between the control cells and that encodes CA1, was strongly associated with breast the cells that were treated with negative control siRNA cancer but with no other tumours. The results suggested (p > 0.05). These results are shown in Fig. 12. that CA1 is a risk factor for breast cancer. Additionally, To identify the details of CA1’s involvement in breast increases in CA1 transcription and translation were cancer tumourigenesis, a PCR array containing 84 differ- detected in breast cancer tissue compared to breast fibro- ent breast cancer-related genes was used to compare the adenoma tissue via western blot and RT-PCR analysis. expression profiles of MCF-7 cells that exhibited down- Benign breast lesions exhibit calcification similar to breast regulated CA1 expression and MCF-7 cells that were cancer tissues. Thus, it was feasible in this study to use treated with AllStars siRNA. Five different breast cancer- fibroadenoma breast tissue as a control when studying related genes exhibited significantly different expression calcification in breast tumours [20]. Using immunohisto- patterns (4-fold or greater changes) in anti-CA1 siRNA- chemistry, extensive expression of CA1 was detected in treated MCF-7 cells. The expression of AR (androgen breast cancer tissues but not in normal tissues. Further- receptor) gene expression was up-regulated, whereas the more, ELISA indicated that the levels of CA1 were signifi- expression of the GSTP1 (glutathione S-transferase pi 1), cantly increased in several of the blood samples that were PTGS2 (prostaglandin-endoperoxide synthase 2), SNAI2 obtained from patients with breast cancer. These results (snail family zinc finger 2) and XBP1 (X-box binding suggested that increased CA1 expression might play a role protein 1) genes was down-regulated in the cells. These in breast cancer tumourigenesis. results are shown in Additional file 1 and Fig. 13a. In the current study, we aimed to identify the role of The results of the PCR array were verified with real- CA1 in mammary calcification by inducing mineralisa- time PCR. Compared with the AllStars siRNA-treated tion of 4T1 cells. Following stimulation with ascorbic MCF-7 cells, the average expression level of AR was acid and β-glycerophosphate, numerous calcium-rich significantly elevated (p < 0.001), and the average expres- deposits were produced in 4T1 cells, and the transcrip- sion level of XBP1 was significantly reduced (p < 0.01) in tion levels of osteogenesis-related genes, including BSP the anti-CA1 siRNA-treated MCF-7 cells. No signifi- and OCN, were significantly increased, indicating that cantly different expression patterns in the GSTP1, the cells underwent calcification and ossification. CA1 ex- PTGS2, or SNAI2 genes were detected in the CA1- pression was significantly increased in the OC-stimulated siRNA cells. The real-time PCR results are shown in cells during calcification. In addition, the OC did not Fig. 13b. significantly increase cell proliferation. Thus, the observed Zheng et al. BMC Cancer (2015) 15:679 Page 12 of 15

Fig. 11 Apoptosis in MCF-7 cells that were treated with anti-CA1 siRNA, as measured using an annexin V cell apoptosis assay. (a) The cells without siRNA treatment were used as controls. (b) The cells treated with Allstars siRNA were used as negative controls. (c) The cells were treated with anti-CA1 siRNA. (d) The result of apoptosis assay is shown in a graph. *indicates p < 0.05, **indicates p < 0.01, ***indicates p < 0.001.

increase in mineralisation was due to a direct effect on The pathogenic pathway of CA1 in breast cancer mineralisation, rather than because of increased cell num- tumourigenesis was investigated using a PCR array that bers. The expression levels of CA1 and calcification were contained 84 different breast cancer-related genes. The significantly decreased when the cells were incubated with results were verified with real-time PCR. The assay acetazolamide, a chemical inhibitor of CA1. These results detected increased expression of the AR gene and de- suggested that increased expression of CA1 promotes the creased expression of the XBP1 gene in MCF-7 cells fol- calcification of breast cancer. We knocked down CA1 lowing anti-CA1 siRNA treatment, suggesting that CA1 expression in MCF-7 cells using siRNA. The down- functions by regulating the expression of AR and XBP1. regulation of CA1 expression significantly stimulated cell AR expression has been reported in approximately apoptosis and cellular migration but did not influence cell 80 % of primary breast cancers [21]. Hu et al. reported proliferation, consistent with what was observed in 4T1 that a significant reduction in breast cancer mortality cells that were subjected to an induction of calcification. was associated with AR expression in patients with ER These results suggested that CA1 expression also mediates (oestrogen receptor) + cancers [22]. Other groups have apoptosis and cell migration, which are two important also reported that AR expression was associated with steps in the process of tumourigenesis. favourable outcomes in patients with ER+ breast cancers Zheng et al. BMC Cancer (2015) 15:679 Page 13 of 15

Fig. 12 Migration of MCF-7 cells that were treated with anti-CA1 siRNA, as measured using a transwell migration assay. (a) The cells without siRNA treatment were used as controls. (b) The cells treated with Allstars siRNA were used as negative controls. (c) The cells were treated with anti-CA1 siRNA. (d) The result of migration measurement is shown in a graph. Original magnification: x4.2. * indicates p < 0.05, **indicates p < 0.01, ***indicates p < 0.001.

[22-24]. Yeh et al. suggested that AR might play a expression of CA1 led to decreased expression of XBP1, positive role in promoting breast cancer progression consistent with previous studies. This finding suggested [25]. Additionally, Venken et al. demonstrated that AR- that CA1 enhances XBP1 gene expression, thereby contrib- knockout mice exhibited reduced bone size and cortical uting to calcification and disrupted apoptosis in breast can- thickness and decreased trabecular bone volume [26]. cer cells. Androgen/AR signalling plays an essential role in bone Our study indicated that the down-regulation of CA1 formation by coordinating the expression of genes that expression significantly stimulated cell apoptosis, but it are associated with phosphate regulation [25-27]. Our did not influence cell proliferation in MCF-7 cells. We results revealed that inhibiting CA1 expression could in- also found that the expression of XBP1 was decreased in crease the expression of AR, suggesting that CA1 could the anti-CA1 siRNA-treated MCF-7 cells. XBP1 can ac- regulate AR expression and might therefore influence tivate the anti-apoptotic BCL2 gene [30]. Thus, it is pos- the progression of microcalcification and other tumouri- sible that the decreased CA1 expression in the cultured genic processes during breast cancer tumourigenesis. breast cancer cells suppressed the anti-apoptotic role of XBP1 is a transcription factor that belongs to the basic BCL2, resulting in increased levels of apoptosis in the region/leucine zipper (bZIP) family [28]. The XBP1 protein treated cells. Apoptosis is not always correlated with cell is expressed in almost 80 % of ER+ breast tumours. The proliferation. For example, Gálfi et al. investigated a set overexpression and splicing of XBP1 have been associated of cell lines to assess their sensitivity to proliferation with poor outcomes in breast cancer patients [29]. XBP1 is inhibition in response to the core histone deacetylase downstream of NFκB, which regulates cell survival via acti- inhibitors butyrate and trichostatin A. They found that vation of the anti-apoptotic BCL2 (B-cell lymphoma 2) there was no correlation between sensitivity to prolifera- gene, as well as several other genes that are associated with tion inhibition and sensitivity to apoptosis induction control of the cell cycle and apoptosis [30]. Additionally, with respect to the histone deacetylase inhibitors [32]. XBP1 signalling increases the expression of Runx2, which Increased migration and apoptosis were observed in modulates calcification in human coronary artery smooth the anti-CA1 siRNA-treated MCF-7 cells, as well as an muscle cells [31]. The present study found that decreased increased expression of AR. Numerous studies have Zheng et al. BMC Cancer (2015) 15:679 Page 14 of 15

enhance bone mineralisation and calcification [36, 37]. Thus, MCF-7 cells could be induced to produce calcification. Carbonic anhydrases catalyse the rapid inter- conversion of carbon dioxide and water into bicarbonate and protons, respectively, which can affect physiological processes in cells by altering pH, proton concentrations, and membrane permeability. In addition, calcium hydroxyapatite has been reported to promote mitogenesis and matrix metalloproteinase expression in varying human breast cancer cell lines [38].

Conclusion This study demonstrated that CA1 expression was increased in samples of breast cancer tissues and blood obtained from patients with breast cancer. CA1 plays important roles in the calcification, apoptosis and migration of tumour cells and contributes to breast cancer tumourigenesis by regulating the expression of AR and XBP1.

Fig. 13 Human Breast Cancer RT2 Profiler™ PCR array analysis of signalling pathways in anti-CA1 siRNA-treated MCF-7 cells. a The expression levels of Additional file four genes, including GSTP1, PTGS2, SNAI2 and XBP1, were down- regulated, whereas the expression of AR was up-regulated in the anti- Additional file 1: Human Breast Cancer RT2 Profiler™ PCR array CA1 siRNA-treated cells. b Real-time PCR analysis of AR, GSTP1, PTGS2, analysis of gene expression patterns in anti-CA1 siRNA-treated SNAI2 and XBP1 expression in MCF-7 cells that were treated with anti- MCF-7 cells. (a) The expression patterns of 84 different genes related to CA1 siRNA. The transcription levels of the target genes were normalised breast cancer tumourigenesis were measured. Fold-change values greater to their corresponding expression levels in the cells that were treated than one indicated a positive finding or up-regulation. Fold-change with Allstars siRNA. Error bars indicate the standard error of the mean. values less than one indicated a negative finding or down-regulation. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p <0.001 Fold-change values greater than 2 are indicated in red; fold-change values less than 0.5 are indicated in blue. (b) A table indicates expression levels of the 84 different genes. Comments with “A” indicate that the average threshold cycle of the gene expression is relatively high (>30) in reported that AR stimulated tumour cell migration, such either the control or the test sample and is reasonably low in the other as during prostate cancer cell invasion and oesophageal sample (< 30). These data indicate that the gene expression is relatively low in one sample and reasonably detected in the other sample, sug- cancer cell migration [33,34]. Apoptosis is disrupted gesting that the actual fold-change value is at least as large as the calcu- during the progression of many types of solid tumours. lated and reported fold-change results. This fold-change result might also In contrast, apoptosis is not always negatively correlated have greater variations if the p value > 0.05. Comments with “B” indicate that the average threshold cycle of the gene expression is relatively high with tumourigenic progression, especially in cases of me- (> 30), indicating that its relative expression level is low in both control tastasis. For example, Termuhlen et al. reported that iso- and test samples, and the p value for the fold-change is either unavail- genic metastatic colon adenocarcinoma cells displayed able or relatively high (p > 0.05). This fold-change result might also have greater variations. Comment with “C” indicates that the average threshold significantly higher levels of staurosporine-induced apop- cycle of the gene expression is either not determined or greater than the tosis than their nonmetastatic counterparts in vitro [35]. defined cut-off value (default 35) in both samples, indicating that its ex- They suggested that the molecular events that were asso- pression was undetected, thus rendering this fold-change result errone- ous and un-interpretable. ciated with the development of a metastatic phenotype sensitised the tumour cells to selecting pro-apoptotic stimuli. Competing interests In the present study, we induced calcification in the The authors declare that they have no competing interests. 4T1 cell line to determine whether a correlation existed Authors’ contributions between CA1 and bio-mineralisation because Cox et al. XC designed the study, coordinated the study and drafted the manuscript. YZ established an excellent in vitro model of mammary performed most of the experiments. BX and YZ performed the genotyping mineralisation using this cell line. However, 4T1 is experiments. YW performed the immunohistochemistry. HG and CL collected RA breast tissue samples, characterised the patients clinically and revised the mouse cell line; therefore, it might not be physiologically manuscript critically. All of the authors read and approved the final manuscript. relevant to human breast cancer. Thus, we also investigated CA1 transcription and translation in MCF-7 cells, Acknowledgements which originated from human breast tumour epithelium. This study was supported by National Natural Science Foundation of China (NTFC) (81171990, 81373218), Shandong Natural Science Foundation MCF-7 cells have been reported to express alkaline (ZR2014HL045) and the Shandong Science and Technology Promotion phosphatase and bone sialoprotein, both of which Program (2014GSF118135 and 2014XGA01011). Zheng et al. BMC Cancer (2015) 15:679 Page 15 of 15

Author details 19. Thouverey C, Strzelecka-Kiliszek A, Balcerzak M, Buchet R, Pikula S. Matrix 1Medical Research Center of Shandong Provincial Qianfoshan Hospital, vesicles originate from apical membrane microvilli of mineralizing Shandong University, Jingshi Road 16766, Jinan, Shandong 250014, P. R. osteoblast-like Saos-2 cells. J Cell Biochem. 2009;106(1):127–38. China. 2People’s Central Hospital of Tengzhou, Tengzhou, Shandong doi:10.1002/jcb.21992. 277500, P. R. China. 20. Nishimura R, Taira N, Sugata S, Takabatake D, Ohsumi S, Takashima S. Suspicious calcifications in benign breast lesions: a radio-pathologic Received: 18 February 2015 Accepted: 7 October 2015 correlation. Breast Cancer. 2011;18(1):33–6. doi:10.1007/s12282-010-0199-4. 21. Brinkmann A, Faber P, van Rooij H, Kuiper G, Ris C, Klaassen P, et al. The human androgen receptor: domain structure, genomic organization and regulation of expression. J Steroid Biochem. 1989;34(1-6):307–10. References 22. Hu R, Dawood S, Holmes M, Collins L, Schnitt S, Cole K, et al. Androgen 1. Morgan M, Cooke M, McCarthy G. Microcalcifications associated with breast receptor expression and breast cancer survival in postmenopausal women. cancer: an epiphenomenon or biologically significant feature of selected Clin Cancer Res. 2011;17(7):1867–74. doi:10.1158/1078-0432.CCR-10-2021. – tumors? J Mammary Gland Biol Neoplasia. 2005;10(2):181 7. 23. Park S, Koo J, Kim M, Park H, Lee J, Lee J, et al. Androgen receptor 2. Frappart L, Boudeulle M, Boumendil J, Lin H, Martinon I, Palayer C, et al. expression is significantly associated with better outcomes in estrogen Structure and composition of microcalcifications in benign and malignant receptor-positive breast cancers. Ann Oncol. 2011;22(8):1755–62. lesions of the breast: study by light microscopy, transmission and scanning doi:10.1093/annonc/mdq678. electron microscopy, microprobe analysis, and X-ray diffraction. Hum Pathol. 24. Agoff S, Swanson P, Linden H, Hawes S, Lawton T. Androgen receptor – 1984;15(9):880 9. expression in estrogen receptor-negative breast cancer. 3. Haka A, Shafer-Peltier K, Fitzmaurice M, Crowe J, Dasari R, Feld M. Identifying Immunohistochemical, clinical, and prognostic associations. Am J Clin microcalcifications in benign and malignant breast lesions by probing Pathol. 2003;120(5):725–31. differences in their chemical composition using Raman spectroscopy. 25. YehS,HuY,WangP,XieC,XuQ,TsaiM,etal.Abnormalmammary – Cancer Res. 2002;62(18):5375 80. gland development and growth retardation in female mice and MCF7 4. Owen TA, Aronow M, Shalhoub V, Barone L, Wilming L, Tassinari M, et al. breast cancer cells lacking androgen receptor. Exp Med. Progressive development of the rat osteoblast phenotype in vitro: reciprocal 2003;198(12):1899–908. relationships in expression of genes associated with osteoblast proliferation 26. Venken K, De Gendt K, Boonen S, Ophoff J, Bouillon R, Swinnen J, et al. Relative and differentiation during formation of the bone extracellular matrix. J Cell impact of androgen and estrogen receptor activation in the effects of androgens – Physiol. 1990;143(3):420 30. on trabecular and cortical bone in growing male mice: a study in the androgen 5. Biltz R, Pellegrino E. The nature of bone carbonate. Clin Orthop Relat Res. receptor knockout mouse model. J Bone Miner Res. 2006;21(4):576–85. – 1977;129:279 92. 27. Kang H, Shyr C, Huang C, Tsai M, Orimo H, Lin P, et al. Altered TNSALP 6. Ripamonti U, Crooks J, Khoali L, Roden L. The induction of bone formation expression and phosphate regulation contribute to reduced mineralization by coral-derived calcium carbonate/hydroxyapatite constructs. Biomaterials. in mice lacking androgen receptor. Mol Cell Biol. 2008;28(24):7354–67. – 2009;30(7):1428 39. doi:10.1128/MCB.00582-08. 7. Damien C, Ricci J, Christel P, Alexander H, Patat JL. Formation of a calcium 28. Reimold A, Etkin A, Clauss I, Perkins A, Friend D, Zhang J, et al. An essential phosphate-rich layer on absorbable calcium carbonate bone graft role in liver development for transcription factor XBP-1. Genes Dev. – substitutes. Calcif Tissue Int. 1994;55:151 8. 2000;14(2):152–7. – 8. Supuran CT. Carbonic anhydrases an overview. Curr Pharm Des. 29. Davies M, Barraclough D, Stewart C, Joyce K, Eccles R, Barraclough R, et al. – 2008;14(7):603 14. Expression and splicing of the unfolded protein response gene XBP-1 are 9. Parissa M, Koorosh A, Nader M. Investigating the application of enzyme significantly associated with clinical outcome of endocrine-treated breast carbonic anhydrase for CO2 sequestration purposes. Ind Eng Chem Res. cancer. Int J Cancer. 2008;123(1):85–8. doi:10.1002/ijc.23479. – 2007;46(3):921 6. 30. Reimold A, Iwakoshi N, Manis J, Vallabhajosyula P, Szomolanyi-Tsuda E, 10. Ramanan R, Kannan K, Sivanesan S. Bio-sequestration of carbon dioxide Gravallese EM, et al. Plasma cell differentiation requires the transcription using carbonic anhydrase enzyme purified from Citrobacter freundii. World factor XBP-1. Nature. 2001;412(6844):300–7. – J Microbiol Biotechnol. 2009;25(6):981 7. 31. Liberman M, Johnson R, Handy D, Loscalzo J, Leopold J. Bone 11. Chang X, Zheng Y, Yang Q, Wang L, Pan J, Xia Y, et al. Carbonic anhydrase I morphogenetic protein-2 activates NADPH oxidase to increase endoplasmic (CA1) is involved in the process of bone formation and is susceptible to reticulum stress and human coronary artery smooth muscle cell ankylosing spondylitis. Arthritis Res Ther. 2012;14(4):R176. calcification. Biochem Biophys Res Commun. 2011;413(3):436–41. doi:10.1186/ar3929. doi:10.1016/j.bbrc.2011.08.114. 12. Chang X, Han J, Zhao Y, Yan X, Sun S, Cui Y. Increased expression of carbonic 32. Gálfi P, Neogrády Z, Csordás A. Apoptosis sensitivity is not correlated with anhydrase I in the synovium of patients with ankylosing spondylitis. BMC sensitivity to proliferation inhibition by the histone deacetylase inhibitors Musculoskelet Disord. 2010;11:279. doi:10.1186/1471-2474-11-279. butyrate and TSA. Cancer Lett. 2002;188(1-2):141–52. 13. Zheng Y, Wang L, Zhang W, Xu H, Chang X. Transgenic mice over- 33. Zarif J, Lamb L, Schulz V, Nollet E, Miranti C. Androgen receptor non-nuclear expressing carbonic anhydrase I showed aggravated joint inflammation regulation of prostate cancer cell invasion mediated by Src and matriptase. and tissue destruction. BMC Musculoskelet Disord. 2012;13:256. Oncotarget. 2015;6(9):6862–76. doi:10.1186/1471-2474-13-256. 34. Zhang Y, Pan T, Zhong X, Cheng C. Androgen receptor promotes 14. Cox R, Hernandez-Santana A, Ramdass S, McMahon G, Harmey JH, Morgan esophageal cancer cell migration and proliferation via matrix MP. Microcalcifications in breast cancer: novel insights into the molecular metalloproteinase 2. Tumour Biol. 2015;28. mechanism and functional consequence of mammary mineralisation. Br J 35. Termuhlen P, Sweeney-Gotsch B, Berman R, Ellis L, Bucana C, Shen Y, et al. – Cancer. 2012;106(3):525 37. doi: 10.1038/bjc.2011.583. Increased apoptosis in metastatic human colonic adenocarcinomas. Cancer ’ 15. Cox R, Jenkinson A, Pohl K, O Brien F, Morgan M. Osteomimicry of Biol Ther. 2002;1(1):58–63. mammary adenocarcinoma cells in vitro; increased expression of bone 36. Delgado-Calle J, Sañudo C, Sánchez-Verde L, García-Renedo R, Arozamena J, matrix proteins and proliferation within a 3D collagen environment. PLoS Riancho J. Epigenetic regulation of alkaline phosphatase in human cells of One. 2012;7(7), e41679. doi:10.1371/journal.pone.0041679. the osteoblastic lineage. Bone. 2011;49(4):830–8. 16. Li Z, Zhang Z, He Z, Tang W, Li T, Zeng Z, et al. A partition-ligation- doi:10.1016/j.bone.2011.06.006. combination-subdivision EM algorithm for haplotype inference with 37. Bellahcène A, Castronovo V. Expression of bone matrix proteins in human multiallelic markers: update of the SHEsis (http://analysis.bio-x.cn). Cell Res. breast cancer: potential roles in microcalcification formation and in the 2009;19(4):519-23. doi:10.1038/cr.2009.33 genesis of bone metastases. Bull Cancer. 1997;84(1):17–24. 17. Hall GE, Kenny AD. Role of carbonic anhydrase in bone resorption 38. Morgan M, Cooke M, Christopherson P, Westfall P, McCarthy G. Calcium inducedby1,25dihydroxyvitaminD3invitro.CalcifTissueInt. hydroxyapatite promotes mitogenesis and matrix metalloproteinase – 1985;37(2):134 42. expression in human breast cancer cell lines. Mol Carcinog. 18. Hall GE, Kenny AD. Role of carbonic anhydrase in bone resorption induced 2001;32(3):111–7. by prostaglandin E2 in vitro. Pharmacology. 1985;30(6):339–47.